Single Image Deraining Using Bilateral Recurrent Network (TIP'2020)
@article{ren2020single,
title={Single image deraining using bilateral recurrent network},
author={Ren, Dongwei and Shang, Wei and Zhu, Pengfei and Hu, Qinghua and Meng, Deyu and Zuo, Wangmeng},
journal={IEEE Transactions on Image Processing},
volume={29},
pages={6852--6863},
year={2020},
publisher={IEEE}
}
Quantitative Result
The metrics are PSNR/SSIM. Both are evaluated on RGB channels.
NOTE:
- Number of training epochs is reduced to 10 for Rain1200 and Rain1400.
| Method | Rain200L | Rain200H | Rain800 | Rain1200 | Rain1400 |
|---|---|---|---|---|---|
| brn | 37.66/0.983 | 28.93/0.906 | 25.93/0.852 | 32.43/0.918 | 30.78/0.918 |
Pretrained models can be downloaded from here
Network Complexity
| Method | Input shape | Flops | Params |
|---|---|---|---|
| BRN | (3, 256, 256) | 197.29GFlops | 375.53k |